What is A/B Testing?

In eCommerce you might test a product page headline, the position of the shipping information, the images in a collection, or the subject line of an abandoned-cart email. Traffic is split between the two versions, and the winner is the one that produces more orders, not more clicks.

How a valid test is built

Start from a question, not from a hunch: decide in advance what you are testing, which metric decides the winner, and how long the test runs. Change one meaningful thing at a time, otherwise you will know that something worked but never what.

Both versions must run at the same time, on comparable traffic, until each has collected enough orders for the difference to mean anything. Low-traffic stores rarely have that volume on a product page, which is why testing usually starts where volume is highest: the home page, the main collection, or email.

Why it matters for an online store in the EU

Advertising costs rise every season, and a store that converts better earns more from the same budget. Testing is the cheapest growth lever you have, because it does not require buying more traffic.

One EU-specific caveat: most testing tools rely on cookies or on storing an identifier in the visitor's browser, which brings them under consent rules. Check that your tool only runs after the visitor has accepted, and that it is listed in your privacy policy. For the legal wording, talk to your own advisor.

Common mistakes

• Stopping the test the moment one version looks ahead, before the result is stable.
• Changing several elements at once, so no conclusion can be attributed to anything.
• Testing cosmetic details on a page that almost nobody visits.
• Running the two versions in different weeks and comparing them as if conditions were identical.
• Declaring a winner on clicks or time on page instead of orders and revenue.